MasoreticaBiblical Manuscript Explorer
What’s new

Version history

v594Current
  • Bug fixReplaced deployment testing with contextual release checks, including precise highlights and a 10% plus 250 ms slowdown threshold.
  • Bug fixFixed local manuscript processing to use its configured Python environment and prepare required tools automatically, with matching publication validation.
v591–v593
  • Bug fixRetained manuscript downloads take turns across batches, so a failed upload does not block later files. Updated retained-product maintenance validation for the reviewed source-ink highlight boundaries. Interrupted manuscript recognition can continue unfinished regions while preserving and reusing verified completed results. Prepared manuscript jobs can run locally through passage highlights and normalized exports while reusing retained recognition and preserving reviewed page structure. Local manuscript preparation connects retained recognition to passage mapping through a fixed local program, with saved evidence and explicit missing-input diagnostics. Added a high-resolution Masoretica icon for websites saved as Mac apps with Safari’s Add to Dock. Fixed a release-check preparation error on Mac. Manuscript font measurements retain diagnostic words without granting them a place in the final text. Manuscript font measurement keeps spacing records consistent when a row is withheld from display. Manuscript overlay checks keep diagnostic font measurements separate from text approved for display. Partial manuscript text overlays preserve font measurements for withheld words without displaying unsupported text. Strengthened publication checks so interrupted verification remains required, and improved validation across Python environments. Preserved the sealed application-release checks when refreshing corpus publication after a site release. Fixed release preparation to find the installed Sites packaging helper in either supported plugin location. Fixed ingestion progress tracking for complete manuscripts with hundreds of images. Improve browser verification and release diagnostics, and preserve exact build packaging when an installed helper is unavailable. Stabilized native browser release checks and restored retained manuscript-product maintenance after its validation tools changed. Manuscript text rendering avoids repeated validation and preserves diagnostic words separately from the displayed text. Manuscript mapping can account for unexplained source writing and withhold unsupported local passage assignments while preserving supported text. Detect incomplete recognition boxes and recover supported passage edges from photographed ink; keep unresolved boundaries explicitly marked. Prevented stalled institutional image requests from blocking other manuscript previews in synoptic view.
  • New featureAdded explicit support for public manuscript image sources with legacy HTTP access and documented IIIF inconsistencies, retaining their original evidence and access limitations. Added secure delivery of retained manuscript images when their institutional source cannot provide a working HTTPS image service. Aleppo’s displayed text now uses Benjamin Denckla’s near-Aleppo edition, including pointed ketiv, documented unwritten-word gaps and edition divisions. The source explanation describes editorial completion where the manuscript is lost, and verse notes retain unresolved readings.
v590
  • New featureEnabled authenticated agents to add manuscripts to the ingestion queue and request processing without opening the ingestion interface, with protection against duplicate requests.
  • Bug fixRestored validation of manuscript publication records after source updates. Restored automatic checks of retained manuscript products after a validation-tool update.
v587–v589
  • Bug fixA resumed ingestion batch can record its validated results after an export failure while preserving the earlier failed state and processing times. Prepare Halper 4’s reader to use source-image bindings from its validated passage index. Register Halper 4’s accepted source-bound passage index for reproducible validation. Preserve validated source-word spacing and repeated-text occurrence choices when exporting manuscript mappings. Preserve recognized word positions and page structure during normalization, fit display text to those positions, and show only the selected passage highlight when the text panel is open.
  • ManuscriptsRe-ingested all four Halper 4 surfaces with fresh recognition, separate raw and normalized text, source-bound passage highlights and reviewed TEI downloads. Damaged passages retain explicit approximate locations. Corrected normalized word placement and display sizing across Halper 4’s four pages, including מראה and ופרקת, while preserving the original recognition and page structure.
v586
  • Bug fixFixed ingestion progress updates failing when retained batch history filled the queue; existing records are preserved. Separated ingestion manuscript and batch records so retained history can grow without filling a single queue record.
v579–v585
  • Bug fixSite updates reuse unchanged validation and application builds, with checks limited to their affected corpus dependencies. Preserve manuscript acceptance when only audit scheduling changes. Manuscript evidence uploads now use the same lightweight private transfer route as downloads. Updated publication checks for the refreshed Halper 22 manuscript evidence. Corrected Halper 22 image links in manuscript features. Published manuscript records now feed the main picker and passage comparisons directly, with source-bound raw recognition and text layers. Publication recovery preserves verified progress and updates the ingestion workspace after each completed publication. Publication checks now accommodate slower reader responses while retaining all content and release checks. Corrected comparison highlights when manuscript recognition uses a resized institutional image, and protected running ingestion workers during upgrade checks. Restored manuscript publication after a website update and added a check for reader compatibility before release. Fixed repeated loading in manuscript comparison, including verses spanning two physical sides. Completeness maintenance can retain larger manuscript histories without stopping at the individual progress-record limit.
  • ManuscriptsRefreshed Halper 22 with source-bound verse highlights, normalized and raw text overlays, downloadable TEI, and corrected descriptions of its pointed Hebrew writing. Halper 22 displays the exact retained scans used for its raw recognition and aligned text. Added Halper 22 with page images, recognition and available text products.
  • New featureThe ingestion workspace updates product completeness with verified manuscript publications and shows retained products by page as ingestion progresses. Older source checks no longer replace newer publication results.
v566–v578
  • Bug fixPreserved clear word spacing and broad source gaps in manuscript text maps without expanding letter spacing. Improved word spacing and preserved paragraph gaps on 51 Or 9905 manuscript text maps. Kept same-line maqaf groups joined while fitting source-sized manuscript text. Text maps measure manuscript letter height and fit word and letter spacing while preserving natural letter shapes. Variants uses the standard button style and shows only the source crop, location and main-text reading. Kept manuscript publishing compatible with the text-map spacing improvements. Reduced memory use when verifying manuscript assets for publication. Made corpus verification resumable in small batches and reused unchanged verification to keep routine updates responsive. Improved processing of small manuscript fragments while keeping unsupported text out of normalized overlays. Updated manuscript publication checks for the corrected fragment text placement. Fixed a validation error interrupting fresh manuscript recognition. Added source-reviewed image preparation for faint manuscript strokes, retaining the exact recognition raster. Rows awaiting text-map placement no longer prevent other rows from using the measured manuscript letter size. Text maps preserve your chosen zoom and keep Hebrew letters and combining marks in the manuscript font in Safari. Enlarged the Daniel text map to match the manuscript lettering and preserved the paragraph gap in Lamentations. Text maps retain manuscript-sized lettering when a crowded row cannot fit, and reuse accepted mapping when only export needs repair. Small manuscript updates now reuse verification of unchanged manuscripts, reducing publication work. Preserve clearly recognized words when damaged manuscript lines leave gaps in recognition. Reduced repeated validation during Textia updates and made release retries reuse verified work. Recover image-verified missed word regions while preserving manuscript line assignments. Keep image-reviewed missed words on their physical manuscript rows. Preserve reviewed word locations while verifying passages only from recognized text. Allow validated manuscript-processing repairs to run independently of website publication. Repair server validation of newly published manuscript page layouts. Restored access to image-bound recognition and text products in newly published manuscripts. Restore access to newly published manuscript assets across Textia hosts. Recognize accepted MAM source references in the biblical manuscript explorer. Published manuscript links now open the main explorer when integrated, and publication checks require manuscript-picker and passage availability. Keep the selected passage, page and language when opening older manuscript links. Manuscript publication now verifies each corpus’s existing explorer and preserves old-link selections through shared redirects.
  • ManuscriptsAdded P.Vindob. H 170 with page images, recognition and available text products. Added P.Vindob. H 170 to manuscript selection and Genesis 48:3–7 comparison.
v552–v565
  • Bug fixCompleted manuscripts can advance from registration to publication in the ingestion workspace without restarting their processing batches. Enabled normalized manuscript text on pages where one work ends and another begins, preserving the accepted passage sequence. Improved manuscript text placement by retaining words across adjoining reviewed regions and allowing natural spacing while preserving valid rows when another row cannot be displayed. Preserved manuscript passage order when loading saved page data, including transitions between works. Complete manuscript inventories now pass consistently from ingestion to catalogue registration. Registered manuscripts preserve surrounding reference context without creating extra passage routes. Or 9905 zoom opens OPenn’s full-resolution JPEGs while keeping smaller previews for page browsing. Restored Or 9905's validated text overlays, including pages spanning a book boundary, in the manuscript's accepted passage order. Fixed missing-verse navigation for registered manuscripts, including Or. 9905, and made comparable raw-recognition lines use consistent typography across Textia. Fixed shared feature-catalogue generation when manuscript adapters use shared reader dependencies. Fixed the manuscript-font catalogue's server rendering boundary so font downloads can be published with the reader fixes. Reused registered manuscript passage indexes during navigation, preventing repeated reference-list rebuilding when a manuscript begins late in the Bible. Corrected normalized manuscript overlays to display their verified manuscript font. Isolated text-map and raw-recognition word direction while preserving source positions and original recognition. Streamed large manuscript downloads in bounded chunks to avoid incomplete ZIP archives. Serve private manuscript downloads without loading the full reader application. Reduce archive compression work so large manuscript downloads can finish within server limits. Whole-manuscript downloads preserve existing ZIP packages and compressed images without recompressing them. Deliver whole manuscripts in standard uncompressed ZIP archives to avoid server compression limits. Restored the manuscript-shaped mercha in the Or 9905 font, text overlays and font downloads. Aligned the Or 9905 text-map font with the manuscript’s recorded writing rows across all 68 available maps. Fixed backwards joined Hebrew words in Safari and made text maps and raw recognition easier to read. Text-map typography now aligns measured font ink with the manuscript’s recorded word regions while preserving passage locations. New manuscript text maps retain verified alignment when published. Added early browser checks, bounded release attempts with retained diagnostics, verified test caching and guarded uploads to prevent avoidable publishing delays.
  • New featureConnected newly registered manuscripts to Special pages and concise bilingual manuscript descriptions. Added manuscript font downloads with source evidence packages, and source-specific fonts for raw HTR and validated text overlays. Added validated manuscript publication with resumable checks bound to the current reader and catalogue updates without rebuilding the application, preserving existing manuscript links. Added source-image comparisons for provisionally reviewed manuscript variants, beginning with Lamentations 2:18 in Or. 9905. Connected Textia corpus catalogues to shared manuscript publication without rebuilding each website.
  • ManuscriptsAdded British Library Or 9905 with source-bound page data and raw recognition.
  • UI improvementUpdated manuscript text overlays with a source-specific font, preserving accepted text and passage locations. Download all available files for a manuscript in one ZIP from the private ingestion workspace, replacing individual file downloads. Variants now opens a simple panel below a normal button, with provisional readings clearly labelled.
v537–v551
  • Bug fixIngestion history retains failed attempts and their measured time while preserving previously accepted pages. Passage locations remain available when a completed text-fitting attempt cannot place an unidentified ink cell, provided the source evidence and reading order still validate. Image-confirmed stains and other non-text marks can be excluded from passage mapping while their original recognition remains preserved. Retained regional recognition can be validated after moving an ingestion checkout, without repeating recognition. Corrected ingestion layout coordinates and consistent spacing checks around uncertain recognized text. Improved ingestion of multilingual manuscript pages and preserved separately measured text components during normalized export. Fixed normalized text export for newly ingested manuscripts. Legacy product checks now advance past pages with missing recognition while retaining their reported blockers. Ingestion progress retains its last update during connection failures and retries without accumulating requests. Ingestion can retain validated normalized text on unaffected rows while preserving failed rows and their passage mapping for follow-up. Preserved valid normalized manuscript exports when browser spacing reaches its allowed limit. Product completeness checks now find retained ingestion batches before manuscripts are registered on the website, keeping unverified captures explicit. Saved manuscript-feature links open their selected collection immediately, with curated results preserved in order. Improved normalized manuscript exports by preserving fixed word positions and retaining rows withheld before browser measurement. Improved publishing reliability when several website updates are ready at once. Ingestion issue links preserve the selected location through sign-in and clear the previous issue when switching manuscripts. Fixed duplicated issue panels during ingestion updates and clarified repeated page counts in timing details. Issue counts no longer show zero while initially loading. Ingestion averages now divide all recorded processing time, including retries, by the manuscript’s actual source page count. Issue images now load without waiting on browser decoding promises, with a bounded wait, retry control and direct source-image link. Keep usable normalized rows when OCR splits words or leaves surplus detections, with image-linked recognition warnings. Retain normalized-row warnings for recognition detections spanning several ink regions.
  • New featureNew Mishnaica registrations use Textia’s shared reader, preserving separate text layers and links to the selected physical page. The shared manuscript reader supports saved TEI downloads and validated normalized text maps with the manuscript’s exact presentation font. Fresh recognition containing main text and marginal notes can continue to passage mapping using its retained writing survey and reviewed page layout. Ingestion can recognize abbreviated marginal writing with uncertain language when one model covers all identified possibilities. Mixed-language manuscript pages can reuse regional recognition while keeping reviewed writing areas and passage mapping separate. The ingestion queue can correct pending language layers after source review while preserving completed work and the earlier plan. Raw manuscript recognition can retain sparse and mixed character order, marking uncertain strings in the viewer and TEI exports. Raw recognition can now preserve sideways and inverted writing at its original position on the scan. Connected source-bound manuscript product checks to Mishnaica and Homerica inventories and added retained-evidence TEI maintenance with explicit per-page blockers. Idle manuscript maintenance now saves its completeness findings in the shared ingestion checklist without changing active requests or publication. The shared ingestion workspace offers downloads of retained ALTO/PAGE XML, linked TEI packages, raw recognition, page layout and available page data across processing batches. Replaced Special pages with searchable manuscript features, keeping curated selections first and making their TEI annotations available. Ingestion issues link to retained manuscript images and affected rows, with passage references and previous attempts available before publication.
  • UI improvementThe ingestion queue and manuscript details show recorded source image totals, distinguish tracked physical surfaces, and clearly label counts not yet recorded. The ingestion workspace now summarizes processing time by step, with individual timings collapsed and averages based on fully measured batches. Ingestion issues now show raw recognition beside source images and explain word-assignment or overlay-fitting failures. Opening another issue keeps earlier images in place instead of jumping the page.
v530–v536
  • UI improvementUnified the Control–Option–R raw HTR view across Textia readers, preserving original recognition at its page positions. The shared ingestion queue retains earlier batches, shows progress across the complete manuscript, and distinguishes remaining mapping from catalogue integration. Start ingestion requests source preparation and continues into processing when ready; the shared queue explains progress and waiting states in English and Hebrew. Unified Special pages across Textia with Masoretica’s collection navigation, image cards and separately labelled cross-corpus references. Ingestion progress lists each remaining surface and preserves completed batch status when retained results are imported.
  • New featureAdded a private Textia ingestion workspace for ordering manuscript queues, recording source preparation and following processing progress across corpora. Raw manuscript recognition supports separate writing regions and flyleaves, preserving each region’s reading direction. Linked the abecedary collection to Cambridge Add. 470.1 in Mishnaica, with each related page labelled by corpus. Preserved complete regional raw recognition while mapping each manuscript text layer independently. Ingestion can acquire OPenn and IIIF images, resume verified downloads, and continue mapping from retained regional recognition. The ingestion workspace shows manuscript product completeness, supports requests to update missing products, and displays recorded processing times. New ingestion exports retain linked TEI evidence. The ingestion workspace can include validated results from earlier batches while preserving current work and manuscript progress. Added reusable catalogue registration and shared page navigation for newly ingested Textia manuscripts, preserving separate text layers and raw recognition. Already-running local ingestion batches can be attached to the queue for live progress, pause controls and restart recovery.
  • Bug fixKept ingestion workspace styles separate from the public manuscript reader. Updated the ingestion queue with current preparation results, progress counts and remaining blockers. Fixed compatibility of manuscript preparation with the production site build. Ingestion progress tracks normalized TEI separately for each text layer and verifies all required manuscript products before registration.
v518–v529
  • Bug fixMade new and corrected manuscript pages available without waiting for a full-site rebuild. Kept the live corpus index on the modular Worker path so new manuscript pages can be published atomically without rebuilding the site. Kept the live corpus index dynamic even when the hosting platform prioritizes static files, so modular manuscript-page releases remain atomic. Made manuscript updates and publishing modular: unchanged HTR stages are reused, global corpus audits are reserved for tool changes and scheduled checks, and the exact tested site build is promoted unchanged. Cleared the website's lint warnings and made future lint checks reject new warnings. Kept framework-specific image linting out of the shared Textia package while preserving warning-free checks. Standardized the Text map across Textia and gave every mapped page one consistent type size, without stretching or clipping individual reference words.
  • UI improvementRemoved redundant explanatory text from the manuscript text-layer selector. Aligned the mobile manuscript reader margins with the selectors and tightened the spacing between the zoom controls. Evened the scan-control spacing and narrowed the Highlight button so the controls stay on one line when space permits. Aligned the scan controls as three evenly spaced units, with verse navigation and Page text kept together at the outer edge. Moved the evenly spaced scan controls into the shared Textia reader so Masoretica, Mishnaica, and future readers use the same compact, aligned layout. Standardized the compact mobile reader gutters across Masoretica and Mishnaica so result text and manuscript areas align with their selectors.
v514–v517
  • UI improvementStabilized the manuscript reader at intermediate widths, keeping page text beside the scan and tightening verse spacing. Moved the reader’s scan controls, source highlight stack, page-structure overlay, and default highlight palette into the shared Textia reader so Masoretica and future corpus sites stay aligned. Restored the brighter yellow highlight control while retaining the normalized, readable manuscript highlights.
  • New featureMasoretica now consumes the shared page-structure and source-highlight primitives directly, with the retained Masoretica page-structure format accepted for compatibility.
  • Bug fixFixed narrow manuscript pages so the page-text drawer sits below the scan instead of covering it. Reissued the responsive manuscript reader fix so compact pages consistently keep the page-text drawer below the scan. Kept the manuscript selector stable at every screen width, including for long manuscript titles. Tightened uncertain manuscript highlights so gaps and stray material between adjacent passages stay outside the selected verse. Kept Halper 31's uncertain verse highlights within the observed writing areas on each page.
v513
  • Bug fixCorrected the embedded orientation of the second Penn CAJS Halper 28 composite photograph. Improved the Spanish Hebrew Bible’s synoptic result card with the Library of Congress’s maximum-size manuscript image. Restored reliable corpus administration and valid general-area manuscript routes in production checks.
  • UI improvementLong manuscript titles now reveal their full text on hover or touch, and MAM reference-text labels are clearer. Opening page text now links every mapped drawer item to its manuscript counterpart on hover; clicking selects it as the active passage. Textia text panels now keep the explored passage text in one readable column. Kept page text to one column, removed stray trailing verse lines, simplified page-text controls, kept the open drawer visible at narrow widths, aligned scan controls with Hebrew RTL reversal, opened direct manuscript views with page text ready, and made manuscript details collapsible. Refined the reader layout with compact manuscript details, balanced spacing, a collapsible details column, and clearer passage and full-page navigation. Improved the finder layout at intermediate widths so the masthead, lookup controls, manuscript selector, and Go action remain readable.
v512
  • Bug fixComparison cards now open the selected manuscript with its own shareable URL, and browser Back returns to the comparison. Normalized text now preserves source-aligned placement on manuscript rows that extend slightly beyond the ordinary column edge. Noisy page edges now retain additional source-supported biblical words instead of dropping them for a marginally cheaper partial alignment. Normalized maqaf text now preserves distinctly separated raw word positions while keeping genuinely joined source runs continuous. Recognized words with reliable line-level geometry now remain on their physical manuscript rows even when finer character boxes are unavailable. Normalized manuscript text now keeps joined maqaf phrases in one source-aligned run and uses measured letter spacing before reducing the page-wide font size. Fresh manuscript retries now rerun only explicitly failed surfaces while rechecking current source and alignment rules. Continuous multi-block pages now share one physical segmentation while retaining independently validated source-bound packages for every block.
  • ManuscriptsPenn CAJS Halper 31 now has a fresh source-bound map for Proverbs 30:27–31:9, including aligned normalized text on the verso, whole-verse highlights, raw HTR, and the visible Masora parva notes.
  • UI improvementMade manuscript highlights paler and more consistent, with clearer text and gentler contrast. Reference navigation now uses the shared Textia control while preserving Masoretica's Book / Chapter / Verse behavior and responsive layout. Normalized verse highlights across parchment tones with a restrained yellow tint that keeps faint writing dark and legible. Added a Corpus selector for moving between the Masoretica Hebrew Bible and the Mishnaica Mishnah while preserving the interface language.
v510–v511
  • UI improvementBrightened verse highlights across manuscripts and improved source-sized HTR lettering without stretching letter shapes. Reconstructed MAM verses now show a concise source credit, with the reconstruction explanation available in the page-text source panel. Reconstructed MAM source notes now stay collapsed until opened, and page text follows each witness's actual column layout.
  • New featureUncertain manuscript word locations are marked Approximate, with broader highlights showing their supported possible area. Manuscript passage navigation and verse highlights can remain available when replacement text cannot be fitted, with broader highlights for approximate locations. The shared Textia reader and manuscript-ingestion pipeline can now preserve physically interrupted or repeated source passages as ordered, independently validated blocks without flattening their source layout. Reconstructed MAM readings now remain available across independently mapped page blocks, while unmapped gaps receive no highlight or page route. Page-text selection and source-bound passage highlights now use shared Textia reader components, so other manuscript projects can reuse the same controls and validated geometry. The shared Textia reader can now load exact manuscript image tiles from remote HTTPS sources without credentials, showing a page only after every tile’s bytes, type, dimensions and placement have been verified; moving to another page cancels the old load.
  • ManuscriptsRefreshed all sixteen Halper 31 verse locations, with wider approximate highlights where uncertain and a normalized text map on the verso. Updated Proverbs 31 on the verso of Penn CAJS Halper 31 with normalized words aligned to the manuscript and complete verse highlights.
  • Bug fixReconstructed manuscript text uses measured font widths to avoid overlapping fixed word positions without moving words between lines. Corrected manuscript text preserves physical lines and word positions, including line breaks across maqaf; verse highlights follow the manuscript independently of display font size. Manuscript text fitting preserves letter proportions and does not use collapsed recognition boxes as whole-word positions. Fresh mapping now gives larger manuscript pages a finite runtime tier derived from their physical rows, source observations, and canonical scope, while retaining the same exact alignment checks. Large manuscript pages now use their complete bounded mapping time before reporting a timeout, without confusing unfinished computation with absent source evidence. Normalized manuscript text now preserves every accepted physical line and interprets sof pasuq against the true verse sequence when a page starts or ends within a verse. Halper 31’s verso now uses the reconstructed MAM reading, including ketiv consonants with qeri pointing in Proverbs 31:4. Halper 31 now shows separate verse highlights, a one-column reconstructed text view, and the reviewed Masora regions without a spurious central loss. Fresh manuscript mapping now recovers mid-verse page edges and adjacent continuations from the source rows, while preserving clear line membership and reserving approximate status for non-isolable page areas. Large multi-column manuscript pages now complete normalized mapping within a bounded page-size budget, while font fitting reuses the same source alignment and preserves physical token order. The MAM reference text now reconstructs documented signs, special letters, and spacing from its Aleppo notes, and uses ketiv consonants with qeri pointing where placement is clear in comparison, manuscript, search, and HTR views. Verse highlights now remain continuous across holes within a physical row and stay separate when neighboring verses have distinct source evidence. The shared Textia reader preserves accepted passage order across books or tractates even when exported lookup keys are sorted alphabetically, while still rejecting changed or missing highlights. Raw and normalized HTR now retain source word positions with one natural-proportion page font; identified-verse highlights are distinct from approximate general page areas. Masoretic manuscript readers now keep the manuscript-style MAM reconstruction as their default while preserving the original MAM attribution on pages pinned to that source.
v509
  • Bug fixThe hidden raw HTR view now follows physical main-text lines at a consistent letter size, excluding stray interlinear detections. Raw HTR words retain their manuscript positions and gaps while sharing a consistent letter size. Improved raw handwriting overlays with source-matched letter sizing and vertical alignment; the hidden keyboard toggle now reads retained recognition across manuscript formats. Text maps preserve natural letter proportions, using smaller type instead of horizontal compression. Missing-page results retain their single-column mobile layout with the shared reader styles.
  • UI improvementThe reader's navigation, page-text drawer, zoom controls and synoptic interface now share reusable components with other Textia projects, preserving Masoretica's manuscript-specific behavior. Borderless yellow highlights brighten the page gently while keeping the ink clear, and matching buttons reflect the highlight color.
  • New featureAdded a hidden keyboard mode for inspecting uncorrected recognition at its original scan positions, initially for both sides of Penn CAJS Halper 31.
v508
  • ManuscriptsRebuilt Penn CAJS Halper 31 with normalized page text, intelligent whole-verse highlights for all sixteen verses, and separate Masora parva and parchment-loss structure outlines.
  • Bug fixRestored Halper 31's verse highlights, normalized text map, and page-structure controls in the manuscript viewer. Enabled accepted occurrence-backed manuscript pages to display their normalized text, complete verse highlights, and page structure without requiring directly legible words in every verse. Made accepted manuscript maps render atomically in the viewer when their verse locations are established by complete occurrence evidence. Kept Halper 31's normalized verse map and complete highlights intact through the final page-order checks.
v507
  • Bug fixManuscript text maps, verse highlights, page text, and verse navigation now stay synchronized, while incomplete mappings remain hidden. Kept evidence-gated manuscript navigation behind a thin routing boundary.
v504–v506
  • Bug fixFuture manuscript additions now use one source-bound word map for both normalized text and verse highlights. Migrated all four Penn CAJS Halper 4 surfaces to one MAM-normalized word map, so Text map and Page structure are available and verse highlights follow the same precise word geometry. Collision-connected compounds now preserve the readable page scale instead of compressing the whole text map. Source-bound manuscript exports now carry the required canonical-sequence and word-capacity evidence automatically, preventing future text-map and highlight migrations from bypassing release validation. Rebuilt all four Halper 4 surfaces with exhaustive MAM-normalized text, source-gated verse navigation, word-aligned highlights, exact split-surface boundaries, comparable full-height column reconstruction, and reviewed outer-right Masorah parva regions.
v503
  • UI improvementMade manuscript descriptions more direct and factual by removing unrelated caveats and internal editorial language. Simplified H15's two-image navigation to concise image links.
  • ManuscriptsRestored H15's complete reconstructed page text in uniform black and corrected line breaks to follow the photographed writing.
  • Bug fixCompleted the H 15 first-side page overlay with tightly bounded biblical text and Masorah parva in all three surviving marginal zones, and removed a misleading conjectural lost-text marker.
v497–v502
  • Bug fixRemoved UXLC editorial note markers from the displayed Leningrad Codex verse text. Corrected the manuscript texts and passage highlights for Joshua 21:36–37, including Bodmer 21’s long reading and manuscript-specific spelling and conjunction differences. Corrected and expanded the Spanish Hebrew Bible’s Joshua 21:36–37 note, documenting its erased wording, extended resh, parentheses, and marginal note about manuscripts without the passage. Manuscript zoom now loads each repository’s highest reviewed scan instead of its fitted preview, including full-resolution LOC, Penn, Haverford, Manchester, IIIF, Internet Archive, and Wikimedia sources; an exhaustive regression check prevents new capped zoom sources. Rewrote the Joshua 21:36–37 manuscript notes to state the two textual sequences and four opening forms directly, and removed the generic transcription boilerplate. Removed the redundant cantillation claim from the revised Joshua 21:36–37 note. Restored the reviewed manuscript highlights for the Reubenite-town passage numbered Joshua 21:36–37 in the interface. Manuscript-specific readings remain primary, BHSA supplies only the fallback text, and the passage note and Etnachta attribution are now clear and concise. Replaced Bodmer 21’s approximate Joshua 21:36–37 boxes with scan-reviewed word-level highlights for the Reubenite-town passage, including its column break. Added an auditable, word-bounded Bodmer 21 highlight layer for the Reubenite-town passage while preserving the original full-page HTR record.
  • ManuscriptsAdded Joshua 21:36–37 readings and direct-sequence notices for every relevant Masoretic witness, with manuscript-specific wording, textual context, and corrected page links. Rebuilt the text maps for both sides of ÖNB P.Vindob. H 15 so the MAM text follows the manuscript’s physical line breaks, including the damaged lower lines.
  • UI improvementH15’s mapped words now appear in solid black; text lost with the papyrus is not drawn as surviving writing.
v495–v496
  • Bug fixImproved ÖNB P.Vindob. H 15 highlights across damaged and split lines, and restored the partially preserved word אולי at the beginning of Genesis 18:24 as an exact fragmentary location. Corrected ÖNB P.Vindob. H 15’s Genesis verse boundaries and damaged-line highlights, including Genesis 18:14–15, 18:19–23, and the surviving incipit of 18:24. Fixed large manuscript comparison presets so they open reliably and expand after the page loads.
  • New featureAdded reconstructed page-layout maps that follow reviewed row layouts and show lower-confidence reference text in pale type, together with a separate view of biblical columns, Masorah, later writing, and physical damage. Split verses can now open each physical part separately, preserve that part in links, show only its securely bounded reference text, and distinguish physically lost words from pale conjectural reconstruction.
  • UI improvementAdded a page-wide explanation that pale gray reconstructions of text physically lost between photographed parts are conjectural display context, not surviving manuscript or HTR evidence. Linked the About page to Textia and explained its role as the home for manuscript-family projects across textual traditions.
v493–v494
  • Bug fixImproved deployment reliability while preserving the existing manuscript and corpus behavior. Improved manuscript search so direct title and shelfmark matches appear first. Added an evidence gate for new HTR verse maps with competing-range checks, repeat-aware word alignment, character-capacity limits, and explicit paratext exclusion.
  • ManuscriptsAdded ÖNB P.Vindob. H 15 with both photographed papyrus sides, 33 reviewed Genesis verse occurrences, and separately selectable halves of Genesis 18:7.
v487–v492
  • Bug fixRepaired Gaster Hebrew MS 710 so every Esther verse highlights in true scroll order, including the overlapping photographs and Haman's sons layout. Rebuilt Gaster Hebrew MS 711 highlighting from reviewed full-resolution text columns and withheld polygons wherever the writing is erased or unreadable. Kept Gaster Hebrew MS 711’s erased middle sections navigable while explicitly preventing false verse highlights there. Repaired Gaster Hebrew MS Add 1 so every Hebrew and Targum occurrence is visible and highlights stay within their reviewed text blocks. Rebuilt Rylands Genizah 32 highlighting from complete-image strip review, preserving the lost recto passage and making all 30 surviving Psalm locations visible. Repaired Manchester Rylands Genizah 2 so all twelve Jeremiah locations now highlight one coherent two-column reading sequence. Clarified that Rylands Genizah 2's twelve verse highlights are bounded review estimates rather than direct OCR matches. Corrected the Hebrew and Saadia Tafsir highlights on Penn CAJS RAR MS 85.126.13a so every retained word follows the manuscript's alternating text order. Rebuilt Penn CAJS RAR MS 85.126.23 so its Hebrew lead-words and Judeo-Arabic spans follow the complete physical text stream. Rebuilt Penn CAJS Halper 22 and restored both selectable physical parts of 2 Kings 5:11 across the page turn. Rebuilt Penn CAJS Halper 25 and restored both selectable physical parts of Zechariah 14:6 across the page turn. Rebuilt Penn CAJS Halper 46 around its ten surviving Hebrew cues and the complete physical Saadia stream.
  • ManuscriptsRebuilt Gaster Hebrew MS Add 19b so every Song of Songs verse has a coherent highlight across the four scroll photographs. Repaired Halper 12a word geometry and its Hebrew–Onkelos page-turn boundaries, including the surviving single-layer endpoints. Rebuilt Manchester Hebrew MS 38 so every accepted word and all 625 physical Psalm occurrences follow its reviewed 12-line page geometry. Rebuilt Rylands Genizah A 59 so all catalogue-bounded Psalm locations are visible while preserving the leaf’s holes and detached material. Manchester Rylands Genizah 1 now follows its reviewed poetic row order with word-bounded highlights and no invented geometry across damaged text.
v477–v486
  • Bug fixPreserved every manuscript route, description, filter, and visibility setting while making catalogue records independently pluggable. Hardened manuscript browsing, search, and comparison against future platform changes. Rebuilt BL Or. 1467 in its word-derived two-column layout, restoring every Hebrew highlight and improving the independent Hebrew and Onkelos geometry across all 257 photographs. Rebuilt Bodley Or. 803’s Hebrew and Onkelos highlights so every mapped Hebrew verse is visible while its exceptional page layouts remain intact. Rebuilt Gaster Hebrew MS Add 17 highlights from all eight text photographs, including the reviewed physical order of the Haman's-sons display. Rebuilt Gaster Hebrew MS Add 18 highlights from all seven text photographs and corrected the physical order of the Haman's-sons display. Rebuilt Gaster Hebrew MS Add 21 highlights from all eight photographs, corrected the Haman's-sons display, and preserved its burned ending without speculative boxes. Restored every Hebrew highlight in Gaster Hebrew MS 2033 while preserving its independent Torah layers, alternating Hebrew-Jonathan haftarot, and displaced Deuteronomy leaves. Restored every Hebrew highlight in Gaster Hebrew MS 322 while preserving its independent Hebrew, Onkelos, and Saadia geometry and the documented missing-folio gap. Restored every Hebrew highlight in Gaster Hebrew MS 323 while preserving its independent Hebrew, Onkelos, and Saadia geometry and reviewed Song of Moses layout. Remapped all 194 surfaces of Gaster Hebrew MS 731 from retained text evidence, restoring every Hebrew highlight while preserving separate low-confidence flags and independent Targum geometry. Recorded the severely burned final Gaster Add 21 photograph as a reviewed unindexable biblical surface, preserving physical navigation without inventing a highlight. Rebuilt Hebrew MS 10 highlights from all eight photographs and corrected the physical order of the Haman's-sons display. Rebuilt Hebrew MS 2 highlights from all seven photographs and removed out-of-order boundary fragments around its Haman's-sons display. Rebuilt Hebrew MS 41 highlights from all ten photographs and corrected the reading order around its Haman's-sons display. Rebuilt Hebrew MS 61 highlights, corrected its final photographed spans, and removed a zero-width OCR artifact. Rebuilt Manchester Hebrew MS 3 highlights across both scribal hands, with reviewed Haman-display and closing-column geometry and honest page-specific confidence. Recorded Hebrew MS 3’s darker second-hand photograph as a reviewed unindexable biblical surface, preserving physical navigation without presenting unsupported highlights. Rebuilt Manchester Hebrew MS 40 highlights from all seven text photographs and corrected the Haman's-sons display geometry. Rebuilt Marshall Or. 51 across all 483 digitized surfaces with independent Hebrew and Onkelos highlights, exact marginal biblical occurrences, corrected Deuteronomy routing, and complete physical navigation. Rebuilt Rylands Genizah 1 highlights in the manuscript's poetic row order, with word-level evidence and no invented geometry for unreadable opening verses. Kept Marshall Or. 51’s two illustrated Temple-plan surfaces in physical navigation without treating them as biblical text. Restored the selectable Targum Onkelos layer for Marshall Or. 51 in manuscript and comparison views. Rebuilt all 689 BL Add MS 9400 surfaces so every accepted text word is accounted for once, multilingual blocks retain their physical order, and implausible verse packing is rejected. Rebuilt Bodley Or. 803 in physical Hebrew–Onkelos reading order, preventing repeated words from scattering one verse across distant columns. Prevented split or repeated OCR lines from scattering one verse across unrelated parts of a page. Rebuilt Manchester Hebrew MS 43 highlighting so every Esther verse is selectable and the Haman's-sons display follows the manuscript's physical order. Rebuilt Manchester Hebrew MS 20 highlighting for every Esther verse and corrected both photographed split-verse joins, including the Haman's-sons display. Rebuilt Manchester Hebrew MS 22 highlighting for every Esther occurrence and mapped the Haman's-sons display in its actual two-part reading order. Rebuilt Manchester Hebrew MS 44 highlighting for every surviving Esther occurrence, separated the repeated words in the Haman's-sons display, and restored the true split of Esther 9:10 across adjoining views. Rebuilt Bodley Or. 699 highlighting as complete page-by-page text streams, eliminating scattered repeated-word matches and withholding 49 unsupported edge routes. Restored precise highlighting for Genesis 50:25–26 in Rylands Samaritan MS 4 and removed unsupported verse locations from its damaged opening. Rebuilt Manchester Samaritan MS 6 highlighting from exact catalogue page boundaries and complete text-block ownership, eliminating scattered matches across all 62 biblical pages. Kept Bodley Or. 699 fol. 33v in physical navigation without assigning it an unsupported biblical highlight. Rebuilt Rylands Genizah A 59’s verso highlights in true reading order and replaced scattered recto matches with two explicitly estimated page-edge locations.
  • UI improvementMade the latest release notes concise and readable.
  • New featureEstablished a rights-aware, reviewable interchange foundation for future Textia witness packages. Added selectable Targum Onkelos with independent word-level highlights throughout the Torah in British Library Add MS 9400.
  • ManuscriptsRebuilt the highlights for Manchester Hebrew MSS 4 and 54 from complete images, including the full-scroll composite and split Esther 9:10.
v467–v476
  • Bug fixVerified and preserved verse lookup, repeated and layered readings, image regions, and complete physical navigation across the manuscript catalogue; corrected isolated navigation and manuscript-identity errors found during review.
  • New featureAdded a bilingual Browse manuscripts catalogue with combined search and scholarly filters for tradition, date, provenance, Masorah, and Babylonian vocalization and cantillation.
v460–v466
  • ManuscriptsRebuilt the complete word-level HTR for Bodley Or. 803 and BL Or. 1473, with independent Hebrew and Targum geometry across every digitized surface. Repaired the complete HTR for Gaster Hebrew MS 322 and 323, with independent word-level Hebrew, Targum Onkelos, and Saadia Tafsir layers in manuscript and synoptic views. Rebuilt the complete HTR for Kennicott 3 and BL Or. 2363 across all 872 photographs, with separately audited Hebrew and Targum Onkelos word geometry.
  • New featureAdded selectable Targum Onkelos for Bodley Or. 803 and Targum Jonathan for BL Or. 1473 in manuscript and synoptic views. Centralized manuscript identifiers in an audited catalog so future text-family adapters can register witnesses consistently. Made Targum Onkelos selectable for both witnesses in individual-manuscript and synoptic views without duplicating manuscript cards or surface data. Made manuscript page routing safer to extend across text families without changing current locations.
  • Bug fixCompacted the new Gaster Hebrew MS 322 and 323 layer indexes to prevent Worker startup failures while preserving their independent Hebrew, Onkelos, and Saadia mappings. Improved the reliability of BL Or. 1473 Hebrew and Targum page navigation while preserving its current locations and complete physical sequence. Hardened BL Or. 1473's complete 357-page physical sequence while preserving every Hebrew and Targum location. Repaired the complete word-level HTR for Gaster Hebrew MSS 2033 and 2034, with separately selectable Hebrew, Targum Onkelos or Jonathan, and Saadia Judeo-Arabic layers wherever each version survives. Rebuilt all 428 surfaces of Bodleian MS. Pococke 285 with complete word-level HTR and separately selectable Hebrew and Saadia Judeo-Arabic geometry in manuscript and synoptic views. Moved BL Or. 2373 page and physical-order routing behind the safer manuscript registry with identical results. Moved BL Or. 4445 page and physical-order routing into the audited manuscript registry with unchanged locations and highlights. Moved BL Or. 9879 page and physical-order routing into the audited manuscript registry with identical results. Moved BL Or. 9880 page and physical-order routing into the audited manuscript registry with identical results. Moved Halper 19 routing into the audited manuscript registry while preserving its exact page results and lazy loading. Moved Halper 2 routing into the audited manuscript registry with unchanged verse and physical-page results. Moved Halper 4 routing into the audited manuscript registry with unchanged verse and physical-page results. Moved Halper 5 and its alternate verse occurrences into the audited manuscript registry with identical results. Moved Halper 6 and its split-verse occurrence routing into the audited manuscript registry with identical results.
v452–v459
  • ManuscriptsAdded four reviewed Austrian National Library fragments—P.Vindob. H 11, H 14, H 133, and H 143—with complete HTR and indexing.
  • UI improvementShortened Hebrew source labels and gave Judeo-Arabic text matching Semitic typography. Simplified the version-history heading by removing its redundant summary sentence.
  • Bug fixStrengthened release checks for duplicate views in the Austrian National Library manuscripts. Corrected the P.Vindob. H 14 Job page ranges and improved its word-level highlights, including verses that share a line. Hardened Berlin 680 verse and page routing with fixed regression fingerprints while preserving every current result. Rebuilt all 114 surfaces of Gaster Hebrew MS Add. 1 and added separate selectable Hebrew and Targum-to-the-Writings word geometry in manuscript and synoptic views. Rebuilt the complete HTR for Gaster Hebrew MS 731 and British Library Or. 1467, separating their Hebrew and Targum text into independently selectable word-level geometry in manuscript and synoptic views.
  • New featureMade version history release-driven: every publishable update now contributes a bilingual note, same-day releases accumulate, and the next active day consolidates the prior day automatically. Prepared manuscript navigation for additional text families while preserving the Berlin manuscript’s existing page results.
v440–v451
  • ManuscriptsAdded Penn CAJS RAR MS 82.258.115, British Library Or. 2365 and Or. 2696, and five short Genizah fragments with multilingual HTR.
  • New featureAdded verse-specific manuscript sharing cards, expanded the About page, streamlined manuscript validation, and advanced the evidence-first Codex Sassoon HTR pilot.
  • UI improvementImproved mobile reference controls, comparison ordering and progressive loading, the default Genesis view, and new-manuscript badges.
  • Bug fixRepaired multilingual HTR for Halper 12a, Gaster 2031, and Reggio 60; prevented whole-corpus rendering exhaustion; and stabilized comparison loading, reference selection, and transient reference-text retries.
v437–v439
  • New featureAdded passage-level Targum tradition choices and a manuscript-based Deuterocanon guide; comparisons now retain every selected tradition.
  • UI improvementImproved responsive tradition controls, Hebrew citations and reference lookup, comparison headings, and the Aleppo Hebrew and Aramaic font.
  • Bug fixRepaired HTR verse geometry, Rösel Targum highlights, Greek rendering, and synoptic tradition navigation.
v434–v436
  • ManuscriptsAdded the Rösel Bible and Library of Congress Targums, including deuterocanonical indexing.
  • Bug fixFixed supplemental-work release regressions, synoptic full-page navigation, HTR sequence logic, and large-witness permalink reliability.
v432–v433
  • Bug fixAudited and repaired Kraken HTR across the corpus, including geometry boundaries and Gaster 731 coverage safeguards.
  • UI improvementMade Special pages count badges responsive on smaller screens.
v426–v431
  • ManuscriptsAdded British Library Harley 1861 and 1528, Add MS 15250 and 15252, Bodleian Canonici Or. 87, and Cambridge Add. 1753.
  • UI improvementGrouped mapped scope into readable ranges, separated citations, and rewrote attribution and workflow credits.
  • Bug fixRepaired Cambridge verse occurrences and strengthened sequence-aware HTR mapping.
v423–v425
  • ManuscriptsAdded BL Add MS 9405–9406, 10455 and 15282, Arch. Selden A. sup. 106, Huntington 234, Harley 5709 and 5715, Oppenheim 14, and Utopia cod. 6.
  • UI improvementMoved mobile transcriptions below scans and fixed the mobile Masorah layout.
v411–v422
  • ManuscriptsAdded Arch. Selden A. 47 and Bodleian Canonici Or. 138, Canon. Or. 91, and Canonici Or. 84.
  • New featureAdded the bilingual About page and the visible release number in the masthead.
  • UI improvementSimplified the About page, improved the sharing card, and adopted MAM in synoptic comparisons.
v410
  • ManuscriptsAdded Bodleian MS Oppenheim 13.
v397–v409
  • ManuscriptsAdded Harley 5506, 5706, 5683, 5710 and 5711; Bodmer 21; BL Add MS 21160; the Cervera Bible; the First Ibn Merwas Bible; and a c. 1300 Spanish Hebrew Bible.
  • UI improvementReplaced synoptic source links with a clearer full-page action.
v378–v396
  • ManuscriptsAdded 16 witnesses from the British Library, CAJS and BLB, including Cod. Reuchlin 10, Add MS 9401–9402, 14760, 4708, 9403, 15451, 26879, 9398, 9399 and 9400, plus Halper 10, 12a, 17 and 30 and a Lamentations Genizah leaf.
v351–v377
  • ManuscriptsAdded 17 CAJS Halper witnesses, BL Add MS 21161, Arundel Or. 2, Genève BGE Ms. heb. 3, Rendel Harris 1, NLI Ms. Heb. 4°7463, and Complutense BH MSS 1.
  • New featureAdded BIMA citations directly to manuscript pages.
  • UI improvementFixed the mobile Explore text layout.
v334–v350
  • ManuscriptsAdded British Library Or. 2542, 1473, 9880, 2543, 1467 and 2363; BLB Cod. Reuchlin 3; and CAJS Halper 5, 8, 9, 14, 15 and 18.
  • UI improvementMade drawer passages flash visibly and adopted MAM for non-Greek witnesses.
  • Bug fixFixed fading highlights and the combined manuscript count.
v311–v333
  • ManuscriptsAdded Rylands Greek P 458, British Library Or. 4445, 2540, 2091 and 9879, CAJS Halper 19, 2 and 4, the Damascus Pentateuch, and Erfurt Bible 3.
  • New featureAdded collapsed manuscript-search facets, reverse exploration from scan highlights, and BIMA zoom for Or. 2091.
  • UI improvementKept BIMA tabs on one row and Explore text scrolling inside its sidebar.
v281–v310
  • ManuscriptsAdded Canonici Or. 137, Reggio 60, Pococke 285, Marshall Or. 1, Canon. Or. 136 and 55, Rylands Genizah 1, 2, 32 and A 59, and Munich Cod. hebr. 2.
  • New featureIntroduced Special pages, manuscript-feature exploration, page and repository citations, citation copy formats, a Genizah filter, and the persistent manuscript intake queue.
  • Bug fixFixed Special pages history, navigation and thumbnails and BIMA Masorah highlights.
v277–v280
  • ManuscriptsAdded full Kraken editions of Laud Or. 324 and Bodley Or. 3.
  • UI improvementRefined citation guidance.
v259–v276
  • ManuscriptsAdded 18 Manchester Hebrew and Gaster witnesses, including five Esther scrolls and fully indexed Hebrew MS 4 and MS 11.
  • New featureAdded bilingual citation guidance.
v245–v258
  • ManuscriptsAdded Manchester Hebrew MS 36, Samaritan MS 6 and Gaster Samaritan MS 1858; Bodleian Canonici Or. 37–40; and Gaster Hebrew Add 13, Add 1, 710 and 711.
  • New featureIntroduced time-limited NEW manuscript badges.
  • Bug fixFixed badge centring and masthead stacking across browsers.
v239–v244
  • ManuscriptsAdded British Library Or. 2373 and completed full-line Alexandrinus HTR.
  • Bug fixStabilized Alexandrinus memory use and cleaned up stale corpus data.
v234–v238
  • ManuscriptsAdded Bodleian Huntington 227, Huntington 367 and Oppenheim 34 with full Kraken indexing.
  • New featurePublished the Alexandrinus full-line HTR checkpoint and kept commentary-only Oppenheim material out of the public viewer.
v204–v233
  • ManuscriptsAdded Manchester Hebrew MS 25 and 38; Gaster 2043, 673, 731, 2033, 2034, 322 and 323; the Berlin Babylonian-cantillation manuscript; and Codex Babylonicus.
  • New featureAdded synoptic filter presets and verse navigation, a manuscript page picker, automatic manuscript counts, availability in reference menus, and approximate Greek verse locations.
  • Bug fixCorrected Aleppo pagination and reconstruction evidence, Berlin Hebrew text and HTR, and Babylonicus multi-line highlights.
v163–v203
  • ManuscriptsAdded Kennicott witnesses 6 and 12–15, Gaster Samaritan Add 1 and trilingual MS 814, and Gaster Hebrew MS 86 and 694.
  • New featureAdded visual highlight correction and review, polygon-point deletion, searchable manuscript selection, computational manuscript details, linked feature analysis, and multi-passage indexing.
  • UI improvementShowed the active verse above results and clarified More info and image-hosting labels.
v147–v162
  • ManuscriptsAdded Kennicott witnesses 3, 9 and 11.
  • New featureAdded a synoptic highlight toggle, pinch-to-zoom scans, refined gesture zoom, and reviewed polygon location suggestions.
  • Bug fixFixed Rylands MS 2 and Sinaiticus highlights and compressed page controls.
v117–v146
  • ManuscriptsAdded Huntington 12, 396, 261 and 591; Bodley Or. 46, 62, 699, 23 and 621; Marshall Or. 3; Laud Or. 174; Rylands Samaritan MS 2 and 4; Canonici Or. 46 and 69; and Gaster Samaritan MS 1133.
  • New featureAdded all-manuscript comparisons, exact Hebrew word search with verse navigation, gap-aware page navigation, and coloured verse controls.
  • UI improvementCompleted Hebrew localization and defaulted first-time Israeli visitors to Hebrew.
v59–v116
  • ManuscriptsAdded and fully mapped Kennicott 1, 2, 4, 5, 6 and 8; Bodley Or. 803 and 804; and Bodley Or. 6.
  • New featureAdded the source-aware page-text drawer, feedback and review workflows, analytics, secure corpus delivery and release verification, linked image-rights information, and full-source releases.
  • UI improvementCompleted Hebrew interface copy, improved manuscript discovery and the picker, and made Read Text more prominent.
  • Bug fixRepaired numerous Kennicott and Sinaiticus highlight, layout and image-tile problems.
v42–v58
  • ManuscriptsAdded Lucianic Rahlfs 82, a fully indexed Samaritan witness, Bodleian Kennicott 3, and Kennicott 7 fragments.
  • New featureAdded grouped manuscript selection, direct image comparison, adaptive high-resolution zoom, shareable comparison links, Keter Aram Tsova typography, and Masoretic app icons.
  • Bug fixFixed Leningrad Genesis 24:1 and Kennicott 3 page mappings.
v1–v41
  • ManuscriptsLaunched with the Leningrad and Aleppo codices, added Sinaiticus and Alexandrinus, and then added Codex Sassoon.
  • New featureBuilt the core verse-to-page finder, HTR line locations, verse highlighting, scan loading, zoom, manuscript comparison, reference-specific page titles, and creator credit.
  • UI improvementSimplified the first result view and tightened the mobile layout and highlight controls.
  • Bug fixCorrected Alexandrinus column and verse boundaries and Aleppo and Sassoon reading-order highlights.

Manuscript features

Curated selections appear first, followed by additional observations.

masoretica: Download TEI annotations

Full-page fields of geometric, interlaced, micrographic, or architectural ornament. This collection includes the sixteen carpet pages of the Leningrad Codex, the three preliminary carpet pages of Kennicott 2, the seven carpet pages of Kennicott 1, the three preliminary carpet pages of British Library Or. 2540, all twenty-three opening and closing carpet pages of Utopia cod. 6, the complete twelve-page painted-arcade series of the Harley Catalan Bible, and the complete four-page opening arcade series of British Library Add MS 15250.

68 observations

Curated text collection

Deuterocanonical and related texts

Some biblical manuscripts—including witnesses in the Masoretic tradition—preserve works outside the Jewish canon beside Scripture or its Aramaic Targum. This guide explains the texts, manuscripts, and scholarship, and updates automatically as occurrences are added.

3 texts · 9 occurrences · 6 manuscripts
About Masoretica

Sources & method.

Manuscript images and metadata come from the holding institutions and digitisation projects listed below. Text and versification data come from UXLC, MAM, BIMA / Corpus Masoreticum, the Codex Sinaiticus Project, the DT-UCPH Samaritan Pentateuch, lxx-morph, and STEPBible. The viewer identifies the source used for each manuscript.

Page mapping and text recognition combine repository indexes, OCR, and transcriptions with kraken, Party, BiblIA, and other specialised HTR models. OpenAI Codex is used in developing the site and in running and reviewing this workflow.

Full source, credit, and licence links appear below.

Sources and licences
UXLC Leningrad index Leningrad scan Leningrad photography credit Aleppo page index Aleppo facsimile Codex Sassoon 1053 page index Sassoon scan Sassoon catalogue Rylands Samaritan scan Rylands Samaritan MS 2 scan Rylands Samaritan MS 4 scan Gaster Samaritan MS 1133 scan Gaster Trilingual Samaritan MS 814 scan Gaster Samaritan MS Add 1 scan Gaster Samaritan MS 1858 scan Bodley Or. 699 Samaritan scan Bodley Or. 23 Isaiah scan Kennicott 1 scan Kennicott 2 scan Kennicott 3 scan Kennicott 4 scan Kennicott 5 scan Kennicott 6 scan Kennicott 7 scan Kennicott 8 scan kraken OCR and HTR toolkit Party multilingual HTR model used for Alexandrinus BiblIA medieval Hebrew model Samaritan kraken model Greek minuscule kraken model Sinaiticus line-segmentation model Miqra according to the Masorah text BIMA / Corpus Masoreticum manuscript transcriptions · CC BY-SA 4.0 Samaritan Pentateuch reference text Alexandrinus facsimile Rahlfs 82 catalogue Rahlfs 82 digitization Vaticanus facsimile Sinaiticus edition Sinaiticus electronic transcription Septuagint reference text Greek–Hebrew verse-number crosswalk Keter Aram Tsova typeface
For scholarly use

How to cite.

If Masoretica has facilitated your research, I would be grateful for a citation.

Preferred citation

Dershowitz, Idan. “Masoretica: Manuscript Explorer.” Accessed 14 October 2026. https://www.masoretica.org/.

For a specific manuscript page, use its Cite button for the complete Masoretica and holding-institution citations. Pages using BIMA also include a BIMA citation, license link, and description of changes.